Food Court Sales Analysis and Prediction Using Machine Learning

Food sales prediction is a time series forecasting task that plays a crucial role in minimizing food wastage and improving business efficiency. This paper presents a comprehensive system for food court sales analysis and demand prediction using machine learning techniques. The proposed system collects transactional data from multiple food stalls and applies statistical and machine learning models including ARIMA, SARIMA, Triple Exponential Smoothing, Support Vector Machines (SVM), and Prophet for accurate sales forecasting. The system incorporates a web-based dashboard that visualizes stall-wise sales, revenue by category, hourly trends, and recent transactions. A Stacked SVM model demonstrated superior prediction accuracy over other methods when evaluated using RMSE and MAPE metrics. The system assists food court managers in inventory management, demand forecasting, staffing optimization, and strategic decision-making